[clang] [HLSL][Matrix] Make matrix single element accessor return a scalar instead of vector (PR #182609)

[Deric C. via cfe-commits]

================
@@ -6,31 +6,27 @@ typedef float float3x3 __attribute__((matrix_type(3,3)));
 void ok() {
     float3x3 A;
 
-   // CHECK:      BinaryOperator {{.*}} 'vector<float, 1>' lvalue '='
-   // CHECK-NEXT: MatrixElementExpr {{.*}} 'vector<float, 1>' lvalue _m12
+   // CHECK:      BinaryOperator {{.*}} 'float' lvalue '='
+   // CHECK-NEXT: MatrixElementExpr {{.*}} 'float' lvalue _m12
    // CHECK-NEXT: DeclRefExpr {{.*}} 'float3x3':'matrix<float, 3, 3>' lvalue Var {{.*}} 'A' 'float3x3':'matrix<float, 3, 3>'
-   // CHECK-NEXT: ImplicitCastExpr {{.*}} 'vector<float, 1>' <VectorSplat>
    // CHECK-NEXT: FloatingLiteral {{.*}} 'float' 3.140000e+00
     A._m12 = 3.14;
 
    // CHECK: VarDecl {{.*}} r 'float' cinit
-   // CHECK-NEXT: ImplicitCastExpr {{.*}} 'float' <HLSLVectorTruncation>
-   // CHECK-NEXT: ImplicitCastExpr {{.*}} 'vector<float, 1>' <LValueToRValue>
-   // CHECK-NEXT: MatrixElementExpr {{.*}}  'vector<float, 1>' lvalue _m00
+   // CHECK-NEXT: ImplicitCastExpr {{.*}} 'float' <LValueToRValue>
----------------
Icohedron wrote:

The implicit cast is for LValue-to-RValue conversion, so I think it's still necessary

https://github.com/llvm/llvm-project/pull/182609